Periodic Reporting for period 2 - FLOWCID (Flow Control for Industrial Design)
Berichtszeitraum: 2023-08-01 bis 2024-07-31
FLOWCID (Flow Control for Industrial Design) was a Marie Skłodowska-Curie Global Individual Fellowships funded by the Research Executive Agency (REA) under MSCA-IF-GF
As part of the FLOWCID project, Prof. Eusebio Valero from the Universidad Politécnica de Madrid (Spain) spent 2 years at Purdue University (USA), collaborating with Prof. Guillermo Paniagua in the Zucrow Lab (https://engineering.purdue.edu/Zucrow) where he conducted a training and research programme. He then completed an additional year in the return phase, transferring the knowledge gained back to UPM.
The main objective of FLOWCID was to develop new methods and tools for controlling flow unsteadiness, particularly the complex non-linear interactions observed in highly detached configurations and unstarting phenomena. To achieve this, FLOWICD proposed combining accurate (high-order) numerical simulations, flow stability and data analysis techniques, along with detailed experimental studies. This approach aimed to model the flow physics involved in Rotating Detonation Engines (RDEs), assess the sensitivity to perturbations, and ultimately define an actuator methodology for controlling these phenomena.
To conclude, this action has provided the fellow with the opportunity to create strong links with American research groups, expand their network of contacts, learn about the educational system of American universities, and improve their knowledge in Fluid Mechanics and experimental facilities. Additionally, the fellow has gained a better understanding of the links between numerical and experimental simulations, facilitating knowledge transfer to and from UPM.
Technically, improvements have been made in the state of the art in numerical simulation, particularly in the use of high-order schemes and their challenges in shock capturing, detached flows, and boundary layer interaction. The interpretation of experimental results, their uncertainties, and comparison with numerical results remain challenging. The creation of an advanced library for feature detection, which merges numerical and experimental data, has improved the accuracy and interpretation of results. Furthermore, new avenues for flow control have been explored, with initial designs indicating that much work remains, especially in unsteady, compressible, and highly detached flows.
The project has had a positive impact on the fellow’s career, who is now internationally recognized in the field of numerical simulation, stability analysis, and flow control.
As part of the FLOWCID project, Prof. Eusebio Valero from the Universidad Politécnica de Madrid (Spain) spent 2 years at Purdue University (USA), collaborating with Prof. Guillermo Paniagua in the Zucrow Lab (https://engineering.purdue.edu/Zucrow) where he conducted a training and research programme. He then completed an additional year in the return phase, transferring the knowledge gained back to UPM.
The main objective of FLOWCID was to develop new methods and tools for controlling flow unsteadiness, particularly the complex non-linear interactions observed in highly detached configurations and unstarting phenomena. To achieve this, FLOWICD proposed combining accurate (high-order) numerical simulations, flow stability and data analysis techniques, along with detailed experimental studies. This approach aimed to model the flow physics involved in Rotating Detonation Engines (RDEs), assess the sensitivity to perturbations, and ultimately define an actuator methodology for controlling these phenomena.
To conclude, this action has provided the fellow with the opportunity to create strong links with American research groups, expand their network of contacts, learn about the educational system of American universities, and improve their knowledge in Fluid Mechanics and experimental facilities. Additionally, the fellow has gained a better understanding of the links between numerical and experimental simulations, facilitating knowledge transfer to and from UPM.
Technically, improvements have been made in the state of the art in numerical simulation, particularly in the use of high-order schemes and their challenges in shock capturing, detached flows, and boundary layer interaction. The interpretation of experimental results, their uncertainties, and comparison with numerical results remain challenging. The creation of an advanced library for feature detection, which merges numerical and experimental data, has improved the accuracy and interpretation of results. Furthermore, new avenues for flow control have been explored, with initial designs indicating that much work remains, especially in unsteady, compressible, and highly detached flows.
The project has had a positive impact on the fellow’s career, who is now internationally recognized in the field of numerical simulation, stability analysis, and flow control.
The project encompasses several work packages (WP) focusing on high-order spectral solvers, modal decomposition techniques, adjoint-based flow control, and low-pressure turbine (LPT) blade simulations.
In WP1, research improved HORSES3D, a high-order spectral element solver using discontinuous Galerkin schemes, by integrating a machine-learning shock sensor based on Gaussian Mixture Models (GMMs). This sensor enhanced shock detection accuracy in supersonic flows, particularly in high-Reynolds-number test cases like the Sedov blast. It also introduced entropy-stable artificial viscosity, improving shock-capturing abilities. The study also reviewed eight modal decomposition techniques like Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD), analyzing their application to fluid dynamics.
WP2 focused on a framework for flow control using continuous adjoint-based analysis for steady and unsteady flows. By solving adjoint problems, the researchers derived sensitivity vector fields, applied surface deformations, and achieved significant drag reduction. Time-averaged deformations proved more effective than time-dependent ones. Additionally, data-driven resolvent analysis was tested for controlling flow unsteadiness behind a 2D cylinder and enhancing heat transfer in turbulent channels, validating its effectiveness.
WP3 involved numerical and experimental studies on a simplified 3D LPT blade model using a wall-mounted bump geometry. The inflow oscillations simulated wake-passing effects, revealing complex flow features like relaminarization and turbulent transitions. Further research into higher Reynolds and Mach numbers showed the need for better numerical resolution to capture vortex formations. The study also applied feature detection algorithms to analyze fluid structures in RDE flow configurations, finding that pulsating base bleed injection efficiently reduces pressure losses in turbine airfoils.
The dissemination efforts of the project yielded eight international peer-reviewed publications and two conference proceedings. A project website was created and regularly updated with key outcomes, while news was also shared on LinkedIn and UPM and Purdue forums.
Collaboration and outreach activities with industry were strong, with regular meetings held at Purdue University. Dissemination extended to academic settings, with students in fluid mechanics classes benefiting from exposure to project findings. Further engagement with the research community occurred through participation in ERCOFTAC and ECCOMAS meetings. Additionally, data produced throughout the project was stored in university repositories and is available upon request, ensuring transparency and open access.
The project also involved collaborative efforts with European projects NextSim and SSECOID, which helped enhance simulation capabilities and allowed for a comparison of results across different platforms and numerical methods.
Lastly, the team is awaiting the outcome of an ERC synergy grant evaluation, which could potentially expand future research activities.
In WP1, research improved HORSES3D, a high-order spectral element solver using discontinuous Galerkin schemes, by integrating a machine-learning shock sensor based on Gaussian Mixture Models (GMMs). This sensor enhanced shock detection accuracy in supersonic flows, particularly in high-Reynolds-number test cases like the Sedov blast. It also introduced entropy-stable artificial viscosity, improving shock-capturing abilities. The study also reviewed eight modal decomposition techniques like Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD), analyzing their application to fluid dynamics.
WP2 focused on a framework for flow control using continuous adjoint-based analysis for steady and unsteady flows. By solving adjoint problems, the researchers derived sensitivity vector fields, applied surface deformations, and achieved significant drag reduction. Time-averaged deformations proved more effective than time-dependent ones. Additionally, data-driven resolvent analysis was tested for controlling flow unsteadiness behind a 2D cylinder and enhancing heat transfer in turbulent channels, validating its effectiveness.
WP3 involved numerical and experimental studies on a simplified 3D LPT blade model using a wall-mounted bump geometry. The inflow oscillations simulated wake-passing effects, revealing complex flow features like relaminarization and turbulent transitions. Further research into higher Reynolds and Mach numbers showed the need for better numerical resolution to capture vortex formations. The study also applied feature detection algorithms to analyze fluid structures in RDE flow configurations, finding that pulsating base bleed injection efficiently reduces pressure losses in turbine airfoils.
The dissemination efforts of the project yielded eight international peer-reviewed publications and two conference proceedings. A project website was created and regularly updated with key outcomes, while news was also shared on LinkedIn and UPM and Purdue forums.
Collaboration and outreach activities with industry were strong, with regular meetings held at Purdue University. Dissemination extended to academic settings, with students in fluid mechanics classes benefiting from exposure to project findings. Further engagement with the research community occurred through participation in ERCOFTAC and ECCOMAS meetings. Additionally, data produced throughout the project was stored in university repositories and is available upon request, ensuring transparency and open access.
The project also involved collaborative efforts with European projects NextSim and SSECOID, which helped enhance simulation capabilities and allowed for a comparison of results across different platforms and numerical methods.
Lastly, the team is awaiting the outcome of an ERC synergy grant evaluation, which could potentially expand future research activities.
From a technical perspective, FLOWCID has contributed in the following ways:
Advancing the development and industrialization of high-order numerical methods through the implementation of robust turbulence models, error estimation techniques, and h/p mesh adaptation strategies.
Enhancing the analysis and understanding of unsteady solutions for canonical problems through detailed numerical simulations and mode decomposition techniques.
Refining continuous and discrete approaches for stability analysis, including turbulence models and complex 3D geometries.
Formulating the sensitivity of flow modes to external perturbations or surface modifications, ultimately leading to the development of new strategies for shape optimization, with applications in flow control.
FLOWCID has consolidated a leading European-American research team in simulation and flow control, with applications in the turbomachinery and aerospace industries. This work has contributed to the design of the next generation of hydrogen-powered engines and aircraft, providing significant environmental and fuel consumption benefits.
In addition, FLOWCID has solidifying Prof. Valero’s position as an international leader in simulation, sensitivity analysis, and flow control. The project also promoted knowledge transfer between America and Europe by facilitating the exchange of junior and senior researchers, enhancing their mobility and positively impacting their career development.
Advancing the development and industrialization of high-order numerical methods through the implementation of robust turbulence models, error estimation techniques, and h/p mesh adaptation strategies.
Enhancing the analysis and understanding of unsteady solutions for canonical problems through detailed numerical simulations and mode decomposition techniques.
Refining continuous and discrete approaches for stability analysis, including turbulence models and complex 3D geometries.
Formulating the sensitivity of flow modes to external perturbations or surface modifications, ultimately leading to the development of new strategies for shape optimization, with applications in flow control.
FLOWCID has consolidated a leading European-American research team in simulation and flow control, with applications in the turbomachinery and aerospace industries. This work has contributed to the design of the next generation of hydrogen-powered engines and aircraft, providing significant environmental and fuel consumption benefits.
In addition, FLOWCID has solidifying Prof. Valero’s position as an international leader in simulation, sensitivity analysis, and flow control. The project also promoted knowledge transfer between America and Europe by facilitating the exchange of junior and senior researchers, enhancing their mobility and positively impacting their career development.